LM7171 Non-Inverting Amplifier Output Weak & Distorted After 32 MHz LPF

[flyingphoenix]

Hi guys, I need an insight, please help me.


I'm building a non-inverting amplifier, where the goal is to amplify a 32 MHz low-passed-input signal. 1kΩ fixed resistor, and a variable resistor with max of 10kΩ.


The original signal is a 100 ps pulse at 500 kHz, connected to a photodiode (PDA10A2). This is then passed through a commercially available 32 MHz low-pass filter (Mini-Circuits – 32 MHz, 50 Ω | SLP-30+), which filters out high frequencies. The output from this filter is connected via a probe to the input of my amplifier circuit.


However, when I observe the output on the oscilloscope, the output signal (purple) differs from the filtered input signal (yellow):

1. The amplitude drops significantly. I expected it to increase due to amplification. Even when the potentiometer is turned to maximum, the output remains low. If the potentiometer is at a value none other than the maximum, no signal appears at all. At this point i dont even know if the pueple waveform is indeed an output or just noise.
2. The waveform shape changes slightly. I was expecting a similar shape to the input, just amplified.

I know that the amplification is limited by the power rails (±5 V), but even when I increased the supply to ±10 V or ±15 V, the output signal still looked the same.


I also made sure the gain is above 2, as stated in the LM7171 datasheet for stability in non-inverting configurations.


I’ve added decoupling capacitors, but they had no noticeable effect.

Please, anyone have any idea? Thanks!

 

[danadak]

Probing is key. The board layout totally unsuitable for this BW. You have a lot of L in your
board. Does TI offer a test PCB for you to look at ?

Bypassing critical, not all caps equal in ESR performance. Using a 50 ohm jig evaluate your
ceramics for bypass performance. I can attest from experience from one vendor to another
big differences.

https://download.tek.com/document/02_ABCs-of-Probes-Primer.pdf

You are using a 10x low C probe right ? Ground clip lead length super short ?

For a high speed OpAmp your absolute value of Rfdbk is too high in my opinion. It forms
a pole with stray C at - input pin = phase shift = reduced phase margin = ringing

OSCON is polymer tant.

Get a copy of TI eval board layout to see just how "tight" the layout needs to be.

An overview of a fast OpAmp -

 

[flyingphoenix]

Hi,

Thanks for your reply.

I used this rigol probe, connected to a rigol oscilloscope and yes, super short ground lead

Rigol PVP2150 150MHz Passive Oscilloscope Probe

Rigol PVP2150, 150MHz, high impedance, passive oscilloscope probe compatible with all Rigol oscilloscopes....

www.rigol-uk.co.uk

From my understanding, your suggestions are:
1. Reduce the Rfeedback to a reasonable value
2. Bypass capacitance

Other than changing from this board to PCB, do you think changing the opamp instead to a relative;y lower speed and lower GBW will help me from these kind of issues?

Since I also have a 1MHz and 5MHz low pass filter filter, so for a gain of 10, I only need a 50MHz opamp. I feel like im spending too much time troubleshooting this very sensitive, very high speed and high GBW op amp and feels like just wanna switch to a much simpler op amp

 

[Galgso]

flyingphoenix your stripboard wiring diagram shows only one pin on the potentiometer connected. Is this a mistake or is that your actual layout? You won't get any results with only one pin connected.

Try a lower frequency initially to make sure your circuit is set up correctly before connecting the 32MHz input signal

 

[flyingphoenix]

flyingphoenix your stripboard wiring diagram shows only one pin on the potentiometer connected. Is this a mistake or is that your actual layout? You won't get any results with only one pin connected.

Hey yes sorry it was the diagram error. I did put two conenctions for the potentiometer. One from wiper to output and another from outer terminal to input

 

[flyingphoenix]

Probing is key. The board layout totally unsuitable for this BW. You have a lot of L in your
board. Does TI offer a test PCB for you to look at ?

Bypassing critical, not all caps equal in ESR performance. Using a 50 ohm jig evaluate your
ceramics for bypass performance. I can attest from experience from one vendor to another
big differences.

https://download.tek.com/document/02_ABCs-of-Probes-Primer.pdf

You are using a 10x low C probe right ? Ground clip lead length super short ?

For a high speed OpAmp your absolute value of Rfdbk is too high in my opinion. It forms
a pole with stray C at - input pin = phase shift = reduced phase margin = ringing

View attachment 150074

OSCON is polymer tant.

Get a copy of TI eval board layout to see just how "tight" the layout needs to be.

An overview of a fast OpAmp -

Hi,

Thanks for your reply.

I used this rigol probe, connected to a rigol oscilloscope and yes, super short ground lead

Rigol PVP2150 150MHz Passive Oscilloscope Probe​

Rigol PVP2150, 150MHz, high impedance, passive oscilloscope probe compatible with all Rigol oscilloscopes....

www.rigol-uk.co.uk

From my understanding, your suggestions are:
1. Reduce the Rfeedback to a reasonable value
2. Bypass capacitance

Other than changing from this board to PCB, do you think changing the opamp instead to a relative;y lower speed and lower GBW will help me from these kind of issues?

Since I also have a 1MHz and 5MHz low pass filter filter, so for a gain of 10, I only need a 50MHz opamp. I feel like im spending too much time troubleshooting this very sensitive, very high speed and high GBW op amp and feels like just wanna switch to a much simpler op amp

 

[Galgso]

In my opinion switching to a lower bandwidth op-amp can only yield worse results. The higher the bandwidth the more fidelity you will get. Have you tried putting in a low frequency signal to make sure your amplifier is set up correctly? 32MHz is a pretty high frequency and your board layout and parasitics will affect the circuit significantly so you want to initially test the amplifier with a low frequency signal to make sure it works at all. If it works at lower frequencies, you will then know that the issue is with your layout. If you don't have a signal generator you can try using a headphone jack from a computer or phone and playing a sine wave audio file.

 

[danadak]

If you go to a lower BW opamp you may not have enough G left to get your target G
which is ?

 

[Galgso]

If you go to a lower BW opamp you may not have enough G left to get your target G
which is ?

Since I also have a 1MHz and 5MHz low pass filter filter, so for a gain of 10, I only need a 50MHz opamp. I feel like im spending too much time troubleshooting this very sensitive, very high speed and high GBW op amp and feels like just wanna switch to a much simpler op amp

 

[flyingphoenix]

If you go to a lower BW opamp you may not have enough G left to get your target G
which is ?

A gain of 5 is enough. So if i use the 5MHz low pass filter, then the GBW is 5x5 =25. And for safe margin, an op amp with GBW of 50MHZ is good enough i believe?

 

[danadak]

But 1 Mhz and higher OpAmps can be senstive to bypassing and stray L and C,
so RF techniques in construction still prevelent.

Some help in construction - https://archive.org/search?query=rf+construction

Some use dead bug construction.....https://hackaday.com/2016/05/04/getting-ugly-dead-bugs-and-going-to-manhattan/

If you use above single sided board useful to minimize ground plane C in a 2 sided PCB.
Tradeoffs, always tradeoffs.

 

[rjenkinsgb]

Impedance matching and termination.

The input from a 50 Ohm filter should be via 50 Ohm cable and terminated with a 50 Ohm resistor to ground at the opamp input.

As others have said, you need a ground plane and decoupling (ceramic caps) to that, with all connections as short as practical. "Dead bug" assembly is the simplest way to test it.

The feedback network should also be low values; more like 500 Ohms to 50 Ohms.

You have not confirmed that the scope probe is set to x10 rather than x1? Also, have you calibrated the probe using the scope's calibration output? You will not get good measurements at high frequencies until the probe is correctly adjusted for the scope.